Large Language Models in Product Design: A Review

Product design is a complex, knowledge-based process that balances customer needs with engineering constraints and market realities. While traditional methods such as Quality Function Deployment (QFD) and machine learning (ML) have improved the design process and solutions, they still struggle with flexibility and a deep semantic understanding of design requirements. Advances in Large Language Models (LLMs), powered by Retrieval-Augmented Generation (RAG) and reinforcement learning from human feedback (RLHF), create new opportunities for intelligent, human-centred support for product design. Following the PRISMA 2020 guidelines, this systematic review analyzes 1,263 studies published between 2023 and 2026 to answer three questions: (1) how LLMs are applied across design stages, (2) how these systems are evaluated, and (3) what gaps remain. We found that LLMs successfully support design requirements analysis, ideation, and documentation. However, current evaluation methods lack standardized benchmarks for design quality. Persistent challenges remain, including hallucinations, weak domain grounding, tool-integration hurdles, and ethical concerns. To guide future research, we introduce a technical maturity map that highlights current achievements and uninvestigated areas in design systems and suggests a roadmap for applying LLMs in product design.